Longitudinal conditional probability of symmetric DNA methylation inArabidopsisplants

DNA replication in high eukaryotes requires the faithful transmission of both the sequence and its methylation status. DNA methyltransferases are not consistently reliable enzymes for accurately restoring methylation in newly synthesised DNA strands when working in isolation. In this study, we evaluated some of the factors that enhance the fidelity of DNA methyltransferases and measure their local and distant impacts on the neighbouring methylation status. A clear understanding of DNA methylation fidelity during replication is important for advances in epigenetic knowledge in general and the potential development of new diagnostics and therapies for epigenetic diseases. We used the Arabidopsis model system because it is moderately methylated, which is an important advantage over animal models since it reduces the chance of a fully-methylated symmetric methylation to virtually nil while providing sufficient occupancy to ensure reliable interpretations. In this study, we employed hairpin bisulfite and nanopore sequencing (1D and 1D 2 ) to test the inter- and intra-strand relationships among the neighbouring 5-methylcytosines (5mC) and evaluate the distribution of hemimethylated sites, as well as the correlations to other cues such as DNA methylation and histone marks. We observed that the presence of a CpG methylation site can predict the occurrence of neighbouring CpG methylation sites, but not necessarily CWG sites, whereas CWG methylation can predict both neighbouring CWG and CpG methylation locations. These predictabilities extend far beyond the size of a nucleosome, and sites 4 kb away still show higher probabilities of methylation compared to the global average. Graphic abstract Graphic abstract Fidelity and longitudinal conditional methylation probability of CpG and CWG DNA methylation in Arabidopsis plants